as the rains. Neither rain, nor wind, nor both, would produce such cliffs in a perfectly homogeneous material such as the tufa seems to be at first glance. Certainly the upper part of the cliff has been longer exposed to the attacks of storms and other atmospheric agencies than the lower part, for the attack began on the upper part as soon as the downward cutting of the stream began to form an incipient canyon, and has continued ever since, while the lateral erosion of the lower part could not have begun until the stream reached the lower level. Consequently in homogeneous material and in the absence of any special attack upon any part of the canyon, the widening of the canyon should have proceeded by equal recession of its walls progressively as the down cutting proceeded thus:
((drawing in field book.))
Instead of as it has, by unequal recession at the bottom thus:
